[HN Gopher] Ubuntu Hoping to Remove Qt 5 Before Ubuntu 26.04 LTS
       ___________________________________________________________________
        
       Ubuntu Hoping to Remove Qt 5 Before Ubuntu 26.04 LTS
        
       Author : sandwichsphinx
       Score  : 65 points
       Date   : 2024-11-01 18:05 UTC (4 hours ago)
        
 (HTM) web link (www.phoronix.com)
 (TXT) w3m dump (www.phoronix.com)
        
       | jftuga wrote:
       | From the article:
       | 
       | "Ubuntu developer Simon Quigley laid out the plans for hoping
       | Ubuntu packages will move from Qt 5 to Qt 6 so that by the time
       | of the Ubuntu 26.04 LTS cycle in early 2026 that the older
       | version of this graphical toolkit can be removed."
        
         | seb1204 wrote:
         | Plans for hoping...
         | 
         | That is no plan at all, it is wait and see.
        
       | 01HNNWZ0MV43FF wrote:
       | Imagine if CLIs routinely died off and had to be rotated out.
       | 
       | Sorry you are running stdout from 2022. You need 2025 stdout to
       | print text.
       | 
       | I guess you can always bundle Qt with your app but the X11 /
       | Wayland split might have complicated that? I just hate it.
       | Viscerally I hate it.
        
         | emptiestplace wrote:
         | Conversely, Windows.
        
           | wongarsu wrote:
           | Both in the sense that Win32 GUIs (introduced in Windows 95)
           | are a bit out of style but are still supported and widely
           | used; and Powershell being a major departure from traditional
           | CLIs.
        
             | emptiestplace wrote:
             | I guess that's what I get for being so terse - I actually
             | meant it as a counterexample. Windows is a good
             | illustration of what happens when you avoid deprecating
             | older systems: you end up with multiple design philosophies
             | and interfaces layered on top of each other, leading to a
             | less cohesive user experience overall.
        
         | duskwuff wrote:
         | Counterpoint: libncurses5 vs libncurses6.
        
         | atomicnumber3 wrote:
         | Part of the benefit here though is that everyone has LOTS of
         | options. Go run Debian or an LTS of something else. App
         | packagers can appimage or flatpack. People can qt5 install from
         | a PPA im sure.
         | 
         | The key thing here is that it lets packagers keep focusing on
         | more impactful things than babysitting projects that
         | can't/won't upgrade, especially when that babysitting ending
         | frees up a lot of brainpower AND is reasonably easily picked up
         | by people who still need that.
        
         | jchw wrote:
         | Qt isn't dying off. They're removing old versions of it from
         | the distribution repositories. (Or well, trying to, anyways.)
         | 
         | Yes, it's neat that you can still compile and run ancient Win32
         | programs on modern Windows, but maintaining all of that
         | compatibility is a burden Microsoft was able to bear. The open
         | source community is still struggling to try to provide a
         | compelling desktop experience without this burden.
         | 
         | Can you run ancient old Linux graphical software? Well, to an
         | extent... Yes. Is it easy? ...No. At least, it's certainly not
         | as easy as installing stuff from your OS vendor directly. As
         | far as Wayland goes, there are no plans to get rid of XWayland
         | any time soon. I'd wager it will remain in the toolkit for
         | decades to come, and if anything, XWayland support keeps
         | getting better rather than worse; just look at what KDE has
         | been doing.
         | 
         | Usually, when I want to try to run something really old I'll
         | use the Docker debian/eol[1] images. Getting X11 apps to work
         | inside of Docker requires a bit of fidgeting, but once you have
         | it working, it's relatively straight-forward to install or
         | build and use software inside of.
         | 
         | The Linux desktop is necessarily going to keep going through
         | large breaking shifts because well, it sucks. It needs to go
         | through breakages. When the new thing also seems to suck, that
         | hints that it might need to continue to break more. It's going
         | to take a while to iron things out, especially because
         | designing systems is hard and sometimes new ideas don't wind up
         | panning out very well. (This has caused a lot of trouble with
         | Wayland where a lot of ideas, indeed, did not pan out super
         | well in practice in my opinion.)
         | 
         | That said, I think we can do much work to improve
         | interoperability with older software, and ensure that it can
         | continue to run in the future. But, it doesn't have to, and
         | really, shouldn't work by keeping everything else from moving.
         | 
         | [1]: https://hub.docker.com/r/debian/eol
        
           | chgs wrote:
           | I find it amusing that there's this constant assetion on this
           | forum that the Linux desktop sucks
           | 
           | Yet many of us have been quite happy with it for decades, and
           | the major problems tend to be with new stuff that seems to be
           | added just to do something different.
        
             | jchw wrote:
             | I have been using Linux as my primary operating system for
             | two decades, and I think the Linux desktop sucks. I
             | strongly disagree that my problems with it are largely the
             | result of "new stuff that seems to be added just to do
             | something different".
             | 
             | When I first started using Linux, there was no kernel
             | modesetting and XFree86 ran as root. I had to manually
             | configure my graphics and display settings by editing the
             | Xorg.conf in nano, or sometimes I could get YaST to do it.
             | Multi-head output was not well-supported, and you often had
             | to restart to reconfigure your outputs. Installing the
             | NVIDIA driver was a straight up pain in the ass. (Some
             | things never change.) X.Org eventually improved and added
             | hotplugging. The Linux desktop stack went through a lot of
             | "new stuff" like new versions of DRM/DRI, the addition of
             | hotplugging, the introduction of KMS and libinput. Dynamic
             | input configuration was added with the `xinput` command.
             | The XInput2 extension was developed to greatly improve
             | touch and pen tablet support. XRandr made it possible to
             | dynamically arrange displays, replacing Xinerama. UI
             | toolkits needed to be rewritten to support scalable
             | rendering so that we could have high DPI support. Most
             | recently, a lot of work and many new protocols across
             | multiple parts of the desktop stack had to be re-hauled
             | from kernel to userland to support explicit synchronization
             | for graphics tasks and buffers. All of this required
             | _major_ shifts, and that broke user 's workflows at some
             | points. Plenty of people were broken by the gradual
             | transition to using libinput, from graphics tablets to
             | Synaptics touchpads, a lot of work had to be redone in
             | order to re-haul the input system. Only recently are there
             | Linux laptops where the out-of-the-box touchpad experience
             | is good enough to rival a MacBook, and it's certainly not
             | all of them.
             | 
             | Don't get me wrong, I loved KDE 3. And despite all of it's
             | flaws, I still prefer Linux greatly over the other options
             | available. But would I go back to using Linux 2.6 and KDE 3
             | and etc. today? _Fuck_ no, we 've improved _so_ many things
             | since then. And yes, no question that at times things have
             | felt like they 've been moving backwards in some regards,
             | but it really isn't for nothing. The fact that modern
             | desktop Linux trivially hotplugs displays with different
             | DPIs and refresh rates (and potentially variable refresh
             | rates) is genuinely something that took a herculean effort.
             | All of that work required massive rewrites. It required
             | kernel changes, graphics driver changes, it required major
             | UI toolkit changes, it required redesigning apps and
             | rewriting components.
             | 
             | And yes, I admit I am not the biggest contributor to open
             | source or anything, my patches are mostly unimportant. But,
             | when I hit something that sucks in the Linux desktop, I do
             | try my best to see if I can't do something about it. I have
             | contributed random little bits here and there to various
             | projects I care about. Most recently, I've been doing work
             | to improve some situations where thumbnails in KDE are
             | suboptimal. There's easily thousands of these little small
             | issues that make Linux worse to use on desktop, we've got
             | plenty of work to do.
             | 
             | And I make it work. Except for work-owned computers, all of
             | my machines run Linux, and I have a fair number of them. I
             | do not have a single box I own personally that boots any
             | other kernel or any other desktop. I know my way around
             | Wine, even enough to make occasional, if small, code
             | contributions. (For example, I made a quick patch when
             | graphics tablets were not working in Wine under XWayland.)
             | So I'm not suggesting that it's unusable, but I can't
             | recommend the Linux desktop to random people. It's been
             | getting closer, but a lot of the reason it's getting closer
             | is because of some of the efforts that I assume you'd
             | complain about.
        
             | andrewprock wrote:
             | Decades? Really? The UX was so bad in 1998 I just went back
             | to windows for another decade.
        
               | chgs wrote:
               | 25 years ago I was quite happy with enlightenment but
               | moved to blackbox/fluxbox. Went to xfce around 2010ish
               | though.
               | 
               | Only issues I can think of that have every affected me
               | were pulseaudio related.
        
               | smallerfish wrote:
               | Windows was so bad in 1999 that I switched to Linux and
               | have been happy with it since.
        
             | yjftsjthsd-h wrote:
             | That's not a contradiction; Linux distros have problems,
             | they just suck differently and probably less than Darwin or
             | NT.
        
             | anvuong wrote:
             | Coming from Windows and Mac, the multiple monitors
             | experience on Linux is terrible, especially when the
             | monitors have different refresh rates or different
             | resolutions. It's getting a bit better with Ubuntu 24.10
             | but still not close to the smooth experience I have on
             | Windows and Mac.
             | 
             | Recently I needed to switch to Wayland due to my monitors
             | setup. Then I also needed to write a small program that
             | capture application screen, which is a stupidly simple
             | thing to do on Windows, Mac, or even on Xorg. For Wayland I
             | had to jump through bunch of stupid hoops just to get it
             | kinda working ...
        
               | viraptor wrote:
               | Having to deal with windows on Windows regularly being
               | placed on a monitor which didn't exist and windows on
               | MacOS restoring to weird places... None of them are
               | great. You get to choose your preferred set of problems.
        
             | LooseMarmoset wrote:
             | The Linux desktop is pretty great, for the most part.
             | 
             | I would say that in addition to a usable desktop, Linux
             | distributions generally also provide a more stable Windows
             | environment than Windows itself does. I have quite a suite
             | of games in Steam and GoG, and I find that many older games
             | that simply cannot be coaxed into running on Windows can be
             | made to run happily on one version of Wine/Proton or
             | another.
             | 
             | There are still some weak areas in desktop Linux if you're
             | running a desktop that isn't Gnome or Plasma - wifi
             | configuration comes to mind immediately, as does sound
             | configuration. Bluetooth can be picky if you don't know to
             | get all the non-free firmware.
             | 
             | But really, we're living in a dream era of Linux desktops,
             | if you happen to have lived through the early days of
             | `startx` and XFree86. Hands up if you ever smoked a monitor
             | coil with a bad X config file...
        
           | acdha wrote:
           | > Yes, it's neat that you can still compile and run ancient
           | Win32 programs on modern Windows, but maintaining all of that
           | compatibility is a burden Microsoft was able to bear. The
           | open source community is still struggling to try to provide a
           | compelling desktop experience without this burden.
           | 
           | I always find it interesting how people talk about this as if
           | Microsoft is just flipping a button rather than spending many
           | millions of dollars on engineers to maintain compatibility.
           | I'm reminded of the systemd arguments where the number of
           | people who are vehement about the old ways being better just
           | never seem to have time to show up to support them. This kind
           | of stuff is expensive and the way to influence the decisions
           | is to show up and chip in.
        
             | 0x457 wrote:
             | Well, that's because deep enterpise users have closed
             | sourced software that cannot be updated (pick any reason,
             | it doesn't matter which one) and they need that
             | compatibility. MS needs to keep them vendor-locked to keep
             | making money off them.
             | 
             | That's not something that happens in Open Source Software
             | world.
        
               | acdha wrote:
               | Yes. My point was just that it's not free but it's weird
               | that done people expect the same level of support from
               | open source projects which don't receive anything like
               | that much money.
        
             | akira2501 wrote:
             | > just never seem to have time to show up to support them.
             | 
             | They're supported just fine. The authors feel no need to
             | bend over backwards to fulfill oddball distribution
             | requirements.
             | 
             | > This kind of stuff is expensive
             | 
             | Yet it all came into existence from open source developers
             | who were paid nothing. Then a bunch of commercial
             | distributions appeared. That's why it's expensive. They're
             | twisting the community for their own profits rather than
             | for broad improvement of the system as a whole.
             | 
             | > to influence the decisions is to show up and chip in.
             | 
             | I don't want to influence things or be required to. I would
             | prefer if people who had "influence" just weren't part of
             | the community. They're exceptionally disruptive and often
             | not focused on users but on their own stature.
        
               | jchw wrote:
               | > The authors feel no need to bend over backwards to
               | fulfill oddball distribution requirements.
               | 
               | This is close to the truth in my opinion, but slightly
               | off in two ways:
               | 
               | - It's not just distributions, it's also other software,
               | especially desktop environments.
               | 
               | - The "oddball requirements" are not so odd and mainly
               | based on real world problems users run into.
               | 
               | There are problems that SysVinit will _never_ try to
               | solve, yet that there is no other obvious place to solve
               | except for pid 1, so systems that continue to run
               | SysVinit will just never have solutions for those
               | problems. And I get it: You don 't care. However, the
               | people who actually work on free software often do. You
               | don't have to love systemd, D-Bus, Polkit, UPower,
               | UDisks, Pipewire, Wayland or any other number of pieces
               | of software, protocols, or specifications, but you have
               | to be in pretty strong denial to not see what lead to
               | them. The entire Linux desktop can't be built on janky
               | shell scripts that parse CLI tool output forever.
               | Eventually, people want to move on and build more
               | complete solutions to problems.
               | 
               | Distributions exert pressure on the ecosystem because
               | they're the ingress for all of the user pain. A vast
               | amount of issues reported go through them first, and then
               | it's up to them to figure out what to do about it. So if
               | you have a problem with anyone, it's probably not
               | actually the distributions themselves, but the kinds of
               | things the users of those distributions are complaining
               | about.
        
               | acdha wrote:
               | > Yet it all came into existence from open source
               | developers who were paid nothing. Then a bunch of
               | commercial distributions appeared. That's why it's
               | expensive. They're twisting the community for their own
               | profits rather than for broad improvement of the system
               | as a whole.
               | 
               | This is bizarrely counter-factual. There are community
               | contributors but an awful lot of open source is
               | contributed by people who are paid to work on it (which,
               | to be clear, is awesome!), even in the case of Debian.
               | 
               | It's also odd in the context of Qt which has been
               | commercially supported since the beginning when Trolltech
               | first released it in 1995. There's always been the idea
               | that if you want long-term support for old versions you
               | should pay for it rather than expecting the open source
               | community to spend time on old code.
               | 
               | > I don't want to influence things or be required to. I
               | would prefer if people who had "influence" just weren't
               | part of the community. They're exceptionally disruptive
               | and often not focused on users but on their own stature.
               | 
               | You're welcome to disrespect them but they have influence
               | because they show up: in open source your voice carries
               | weight in proportion to your contributions. It's also
               | inaccurate to say they're not focused on users: you might
               | not think you need a given feature but you're not the
               | only user out there. Alternatives are there but most
               | people aren't using them because they don't feel the need
               | to switch as strongly as you might.
        
               | pessimizer wrote:
               | > in open source your voice carries weight in proportion
               | to your contributions.
               | 
               | And your contributions grow in proportion to your income,
               | and a lot of people get their income from companies who
               | have other than the best interests of Linux users at
               | heart, and rather want to steer it in a direction that is
               | most profitable for themselves.
               | 
               | That's certainly fine for open source, but that's not
               | fine for these distros as a whole, at least the ones that
               | portray themselves as a service to the public.
               | 
               | > It's also inaccurate to say they're not focused on
               | users: you might not think you need a given feature but
               | you're not the only user out there.
               | 
               | If your voice carries weight in proportion to your
               | contributions, users are not part of that equation. Users
               | enter the equation if they're the users that are
               | desirable for the people who are paying contributors.
               | 
               | > Alternatives are there but most people aren't using
               | them because they don't feel the need to switch as
               | strongly as you might.
               | 
               | And because people who insist that Linux move in a
               | particular direction immediately start breaking things
               | that used to work, and making their new thing a
               | dependency. Then the alternatives have to spend all their
               | time shimming that stuff, and the best they can hope for
               | is to almost keep up. The prize for becoming an
               | alternative is having to become an expert on the new
               | thing that you disapproved of, or fall behind. You never
               | get time to work on more parsimonious solutions to the
               | problems the new thing claimed to solve.
        
           | akira2501 wrote:
           | > The open source community is still struggling to try to
           | provide a compelling desktop experience without this burden.
           | 
           | What exactly is a "compelling desktop experience?" It seems
           | to me that when people say this they really mean "win the
           | popularity contest against Microsoft." I'm not sure that's at
           | all a worthwhile or laudable goal.
           | 
           | > Is it easy? ...No.
           | 
           | It's incredibly easy. It just depends on what distribution
           | you use. For example, on Gentoo, this is not at all a
           | problem.
           | 
           | > When the new thing also seems to suck,
           | 
           | This is a hint that you got something wrong. Remember when
           | desktops used to be configurable? I guess letting the user
           | control their own environment stopped being "compelling" when
           | Microsoft decided so. Do we have to follow suit?
        
             | jchw wrote:
             | > What exactly is a "compelling desktop experience?" It
             | seems to me that when people say this they really mean "win
             | the popularity contest against Microsoft." I'm not sure
             | that's at all a worthwhile or laudable goal.
             | 
             | I don't care about Microsoft or Apple. I don't like Windows
             | and I don't like macOS. I care about _my_ desktop. The one
             | that I am typing in. I really, _really_ hope I don 't
             | _actually_ have to explain all of the problems with the
             | Linux desktop experience right now, I don 't have the
             | energy. I've helped a lot of people try to make Linux work
             | for them and it's a soul-crushing experience every time
             | because I often have to explain that there is in fact, no
             | great options for them right now because shit is simply
             | broken. The X11 and Wayland situation is a perfect example:
             | "Oh no problem. You can just choose between your windows
             | scaling properly and your high refresh rate monitor
             | working, and being able to actually use screensharing
             | during meetings." Sure it's getting better, but we've got a
             | lot of work to go, there's no sugarcoating it.
             | 
             | > It's incredibly easy. It just depends on what
             | distribution you use. For example, on Gentoo, this is not
             | at all a problem.
             | 
             | Look, if the person has to be proficient enough to be able
             | to _install and operate Gentoo as their primary desktop
             | operating system_ , I'm pretty sure they do not need my
             | help when it comes to doing virtually anything with Linux.
             | It's also pretty easy to run old versions of software on
             | NixOS but I'd argue this is basically cheating.
             | 
             | What's hard to do is take any of the most popular distros
             | and run something from 1999. (I'm pretty sure that's
             | actually generally hard to do on Gentoo as well, but
             | whatever.) That's what I'm really talking about.
             | 
             | > This is a hint that you got something wrong. Remember
             | when desktops used to be configurable? I guess letting the
             | user control their own environment stopped being
             | "compelling" when Microsoft decided so. Do we have to
             | follow suit?
             | 
             | What exactly about modern KDE isn't configurable enough? I
             | use both KDE and SwayWM as daily driver desktop setups and
             | never felt hindered much by the available configuration
             | options. If anything, KDE offers so much customization that
             | it's probably sometimes detrimental.
             | 
             | Configuration options come at a cost, and that's exactly
             | why modern software has less of it. We want higher quality
             | software, but it's hard to make software more robust when
             | you have trillions of code paths and most of them are
             | almost completely unused. This stuff adds up. In a similar
             | vein, I hate to see Linux lose support for old hardware,
             | but if something's been broken in mainline for five years
             | straight, I'm pretty sure they have every reason to assume
             | it's not being used by anyone.
             | 
             | What I'm really talking about with "the new thing also
             | sucks" is more reflecting on the growing pains of things
             | like Wayland and PulseAudio (and a bit with Pipewire though
             | less so.) I am of course not saying that the same thing
             | doesn't happen with desktop environments, personally I
             | don't like GNOME 3, but also in the same vein, it'd be hard
             | to argue GNOME 3 has been getting _worse_ this whole time,
             | it offers much more customization and is a lot more robust
             | than it was when it first launched.
        
         | pjmlp wrote:
         | Image if stdout still required a physical UNIX teletype from
         | 1970's actually work.
        
           | neilv wrote:
           | Did it ever, or was that the tty abstraction rather than
           | stdout?
        
         | serbuvlad wrote:
         | I hate dynamic linking and this is one of the reasons. APIs
         | should be small, self-contained, not break backwards
         | compatibility (except for security reasons). Libraries provide
         | much more extensive features and, as such, need to change a lot
         | more often.
         | 
         | APIs are: Linux system calls, the std{in,out,err} convention,
         | termcap/terminfo, X/Wayland, D-Bus, PulseAudio/PipeWire etc.
         | 
         | These change very rarely, and when they do they generally offer
         | compatibility layers (XWayland, pipewire-pulse).
         | 
         | But dynamic linking makes every single library an API that can
         | break!
        
         | zamadatix wrote:
         | That's why I stick to using sh. Wait no, bash... but make sure
         | it's got all of the features from the 5.2 release... unless
         | it's when I'm using zsh in which case we'll need to change some
         | syntax around. For interactive use the program has an ncurses 6
         | interface, if you want to manage it that way, so make sure you
         | aren't still back on ncurses 5 and have an xterm-256color
         | terminal!
         | 
         | Jokes aside, stdout is more like the X11/Wayland comment than
         | anything... and even then it's been what, >30 years with one
         | backwards compatible swap there?
        
           | yjftsjthsd-h wrote:
           | Where's the joke? I _do_ in fact target somewhat oldish POSIX
           | sh when writing shell scripts; I want to write a script once
           | and not have to worry about where it will run, and sh is
           | basically universal.
        
             | zamadatix wrote:
             | The joke was the over the top-ness in the example of how
             | CLI solutions have lacked forward compatibility as often as
             | GUI solutions. The latter parts about how both CLI and GUI
             | have decent examples of long term backwards compatibility
             | were serious.
        
               | yjftsjthsd-h wrote:
               | I don't think it works that way either, though. AIUI the
               | problem is that QT5 apps don't automatically work on QT6
               | (otherwise this would be a non-problem; Ubuntu would just
               | ship QT6 alone and everything would work). But none of
               | the shell examples given have that problem; AFAIK
               | virtually every script that worked on BASH 4 works on
               | BASH 5 - as does virtually every script that worked on
               | any other older version of BASH, and every script that
               | targets plain sh. If you wanted to make that argument,
               | the ncurses suggestion by
               | https://news.ycombinator.com/item?id=42020272 is probably
               | more compelling than the shell itself, which is
               | _amazingly_ backwards compatible.
        
               | zamadatix wrote:
               | Sure, I'll give the particular example of bash compared
               | to sh is probably a poor example to put focus on in that
               | it's one that doesn't demonstrate every point just like
               | SDL 1.2 to 2.0 wouldn't due to SDL_Compat (leaving some
               | room for interpreted vs compiled differences to not count
               | otherwise we'd have to restrict conversation to scripted
               | GUIs instead of QT or compiled shell commands instead of
               | scripts). The other example of ncurses 6 vs 5 is better
               | as it's a compiled interface where the ABI broke but that
               | was already in the comment. Or take the other example zsh
               | vs bash on how there are often changes that take some
               | porting work but not much.
               | 
               | More than each of the specific examples, the main point
               | was meant to be more around things like stdout being
               | comparable to display server protocols which do behave
               | about the same in compatibility. Specific apps end up
               | doing their own thing with compatibility regardless which
               | interface they output on. One of the examples I chose was
               | a poor example for that and seems to have distracted from
               | that point, mea culpa. Another commenter noted Python
               | might have been a better example of scripting
               | compatibility over time.
        
               | Brian_K_White wrote:
               | Doesn't work.
               | 
               | A 30 year old sh script still works.
               | 
               | A 1 year old qt5 app or python script does not.
        
               | zamadatix wrote:
               | You're right there, I should have used a Python on the
               | CLI side as the example. Bash was a poor choice as it
               | doesn't work with all of the points in one example. GUIs
               | also vary on their out of the box compatibility, QT5 to
               | QT6 is similar to Python in that that doesn't (without
               | some work) just work though.
        
       | robomartin wrote:
       | I have always avoided Qt for any projects because of the
       | incomprehensible licensing model that could very easily create a
       | legal minefield for your product.
       | 
       | No. Thanks.
        
         | jcelerier wrote:
         | It's just LGPL if you make any standard desktop app. Like you
         | can make pretty much the entirety of KDE with the LGPL parts of
         | Qt.
        
       | foresto wrote:
       | I think this is reasonable. Plasma 6 recently appeared in Debian
       | Experimental, Qt 6 is already in the Stable branch, and a new
       | Debian release seems likely in the coming year. I expect the
       | folks managing Ubuntu just want to make sure their non-Debian
       | stuff is ready for the upstream change.
        
       | mmsc wrote:
       | Forcing modernisation of some - (probably) especially gui -
       | applications is overall a positive for the platform as it also
       | identifies projects which are under-maintained and allow for
       | overall modernisation / forking / maintenance that is needed.
       | 
       | I wouldn't expect to (easily) run php5 code on Ubuntu 26.04,
       | either. Applications that refuse to modernise can build qt5
       | themselves.
        
         | yjftsjthsd-h wrote:
         | > and allow for overall modernisation / forking / maintenance
         | that is needed.
         | 
         | Or, it allows for pointless churn, extra work, and breakage of
         | apps that were still functional until someone yanked their
         | dependencies. This kind of thing is why software rot is even a
         | thing and people remarking that Win32 Is The Only Stable ABI on
         | Linux ( https://blog.hiler.eu/win32-the-only-stable-abi/ ).
        
           | mmsc wrote:
           | Ubuntu does not claim they have a stable ABI between major
           | version releases.
        
             | yjftsjthsd-h wrote:
             | Sure? Acknowledging a shortcoming of an OS doesn't remove
             | it.
        
       | kev009 wrote:
       | I think it makes sense to try and cycle the majority of the
       | dependency network to the new versions but I also think it
       | probably makes sense to let the old toolkits live on indefinitely
       | (maybe strategically drop the webkit stuff everything else should
       | be low maintenance) for a variety of reasons, particularly old
       | commercial software.
       | 
       | It is comical in some sense that a Motif app of yore would've
       | been a great "investment" in terms of still functioning now and
       | for the foreseeable future.. sort of like Win32/MFC.
        
         | mrweasel wrote:
         | > Motif app of yore would've been a great "investment" ... sort
         | of like Win32/MFC.
         | 
         | I've slowly started getting back to desktop Linux, and honestly
         | Motif (mwm) is fine for my needs. Maybe fvwm would be slightly
         | better.
         | 
         | If I where to do my career over, I'd consider going for C++ and
         | Win32, it seems like a stable performant platform (I get that
         | there's other problems on modern Windows).
        
           | exe34 wrote:
           | wine is the stable abi on Linux.
        
           | rjsw wrote:
           | I'm using mwm right now, and a motif build of emacs.
        
       | happosai wrote:
       | Canonical, the company behind Ubuntu, could easily port the
       | remaining Qt5 apps to Qt6 by the deadline. But they rather wait
       | for the community to do it for them.
        
         | NGRhodes wrote:
         | I suspect that means waiting for Debian to do it.
        
         | eptcyka wrote:
         | Porting them out of tree would be worse than doing nothing.
        
         | rlpb wrote:
         | I think you overestimate Canonical's size.
         | 
         | What _is_ happening though is really long support for older
         | releases being provided by Canonical so you will be able to use
         | older Qt 5 based apps on Ubuntu, security supported, until
         | 2034, on Canonical 's dime. With container technology
         | (something that Canonical significantly funded, with Docker
         | originally being based on Canonical-funded LXC) you'll be able
         | to run those older apps on a newer base OS including future
         | ones, too.
         | 
         | Those are huge contributions to the _user_ community you seem
         | to be discounting.
         | 
         | Disclosure: I work for Canonical. I don't speak for Canonical.
         | But what I say above are simple verifiable and falsifiable
         | facts and not really a matter of opinion.
        
         | pkaye wrote:
         | What about the other distros? Are they waiting on the community
         | to do it for them?
        
       | larrik wrote:
       | On the one hand, yes.
       | 
       | On the other, why does Ubuntu always do their craziest changes in
       | the LTS? The LTS should boring, and the one _after_ it should be
       | the wild one. (That may not be what the plan is here, but it does
       | sound like it)
        
         | mort96 wrote:
         | I'm guessing it's because everything that ships in the LTS must
         | be maintained for 12 years. I understand Canonical's desire to
         | not still be maintainig Qt 5 in 2038.
         | 
         | Tho the ideal time to make this change would be in 25.10.
        
           | chris_wot wrote:
           | Perhaps they need to reconsider how an LTS works?
        
             | baq wrote:
             | In enterprise settings 12 years is not quite enough. There
             | are racks of boxen set up with custom software targeting a
             | particular distro version and there is no money or value in
             | making changes to make it compatible with something newer.
             | 
             | Hardware gets upgraded on 5-7 year cycles, but the
             | software... isn't happening.
        
             | mort96 wrote:
             | The value of LTS, and the reason companies spend tonnes of
             | money on support contracts for those 12 years, is exactly
             | this though: the set of libraries shipped in 26.04 will
             | still be supported 12 years later without major backward
             | incompatible changes. The very reason people would pay to
             | keep using 26.04 in the mid 2030s and beyond would be that
             | the software they built or bought in 2026 keeps working,
             | typically without even having to recompile.
             | 
             | You may argue that those customers should just update their
             | software to use new versions of libraries or frameworks or
             | replace discontinued libraries with new ones, or pester
             | their vendor to do those things, or pay a bunch of money to
             | consultancy companies to do those things; and you may well
             | be right, but as long as those companies won't do that and
             | are willing to instead pay Canonical, why wouldn't
             | Canonical accept the money?
        
               | secabeen wrote:
               | Vendors don't always want to do this. I've seen 6-figure
               | instruments stop working after a major update, and when
               | we go to the vendor for an updated version of the
               | software, their response is "we don't support that
               | hardware anymore; buy a new instrument."
        
           | spockz wrote:
           | I think GP is intending to have it work as follows:
           | 
           | If canonical would like to remove QT5 from LTS vN, they need
           | to remove it in the first non-LTS release after LTS v(N-1).
           | That way there is a lot of time before it to adopt which
           | increases the chances of it actually happening at vN.
        
       | butz wrote:
       | I am hoping KeepassXC will move to Qt6 soon.
        
       | dgfitz wrote:
       | Kind of sounds like Canonical and Qt are in cahoots, in a similar
       | way that Qt and openembedded are.
       | 
       | I sure don't like Canonical.
        
         | mort96 wrote:
         | I sure as hell would prefer to not have to maintain Qt 5 for 13
         | more years if I could avoid it. I don't think this has anything
         | to do with "being in cahoots with the Qt company", I think it
         | has everything to do with "backporting fixes from a recent Qt 6
         | version to an older Qt 6 version is easier than backporting
         | fixes from Qt 6 to Qt 5".
        
       | m463 wrote:
       | clickbaity title. Makes it sound like Qt being removed.
       | 
       | Better might be:
       | 
       | Ubuntu Hoping to Remove Qt 5 in favor of Qt 6 Before Ubuntu 26.04
       | LTS
       | 
       | or maybe:
       | 
       | Ubuntu will try moving from Qt 5 to Qt 6 by Ubuntu 26.04 LTS
        
       ___________________________________________________________________
       (page generated 2024-11-01 23:00 UTC)